Say it. Moro does the rest
Tap, speak, and it lands as finished text. Fillers gone, punctuation fixed, layout done. Translate, rewrite, note, open apps, all with one key.
Four questions worth asking
Moro is not dictation with an AI label slapped on. The four questions that actually matter: is input faster, does it work across apps, does privacy hold up, and what does it cost over time. Here’s where we stand.
Voice runs ~180 chars/min against ~60 typing — before the cleanup and punctuation it saves. (estimate)
System-wide across most standard macOS text fields: WeChat, Lark, Notion, Cursor, Mail, Notes, and more. Some custom fields may be limited.
Flip to local-only: transcription and processing stay on your Mac, 0 content leaves; password fields always skipped.
Buy the software once, no monthly fees; local models cost ¥0, cloud runs on your own key.
One key. Every intent
Beta · evolving fastNo keymap to memorize, no mode to pick first. Just say it, and Moro works out what you want.
“Turn this feedback into a client-ready reply.”
client-ready text, in place
“Any clause here that works against me?”
answers on your selection
“Note: file this week’s expenses before I leave.”
saved as a tidy to-do
“Open Calendar so I can check next Wednesday.”
opens the right app
Beta“Help me find last month’s rental contract.”
results card pops up; one click reveals in Finder
Beta“Switch to local-only processing.”
changes Moro’s own setting
BetaAction intents like Open, Find and Settings are still maturing. When unsure, Moro falls back to plain dictation rather than acting blindly, and it asks before anything sensitive. It remembers the corrections you make.
Everything you need to get the words right
Voice to text
3× faster than typing. Fillers removed, punctuation and numbers fixed. What lands is ready to send.
Quick notes
Jot by voice, filed as tidy to-dos. Say “the weekly report is in” to check it off.
Triple-Space translate
Place the cursor in a paragraph and tap Space three times—no selection needed. It writes back in place, or floats beside read-only text.
Rewrite & polish
⌥R rewrites, ⌥S polishes. The same line reworded — clearer and more natural — replaced in place, and you can undo it.
Quick chat
Ask in a flash. Attach your selection, a file, or a screenshot, and use the answer right away.
Type instead
Can’t speak? Double-tap ⌘ to type. Same intents.
Not just dictation. A full input chain.
Most voice tools are judged only by transcription accuracy. Moro connects what you said, where you are, what you meant, which model to use, and where the result should land.
Input layer
Hold a shortcut to speak, or type when speaking is awkward. Fewer entry points, less muscle-memory cost.
Intent layer
Classifies write, translate, rewrite, note, ask, open, find and settings. If unsure, it falls back to plain input.
Model layer
Cloud handles higher-quality output, local keeps things private. Each feature can route to different models and compatible endpoints.
Output layer
Results land in the active app; notes become local cards; terms, style and stats stay on your Mac where possible.
Time saved, made visible
Speaking is usually faster than typing. Moro converts each input into minutes saved and keeps the tally on your in-app stats page — usage, time saved and per-model performance, recorded on your Mac only.
Estimated for Chinese at ~60 chars/min typing vs ~180 speaking. The in-app tally uses a more conservative formula, so real numbers may be lower.
Voice ~180 chars/min vs typing ~60 — a 3× floor, assuming these inputs go by voice. The cleanup and punctuation it saves aren't even counted.
No new workflow. Moro works across most everyday text fields
WeChat
Lark
WeCom
Notion
Cursor
VS Code
Chrome
Safari
Mail
Notes
Figma
Messages
Telegram
Claude
WhatsApp
DiscordWorks with most standard macOS text fields; write-back and selection actions may be limited in some Electron apps, web editors, or custom fields.
Notes, typing and autocomplete
Catch a thought
the moment it strikes.
Tap ⌥D and just say it. Moro files it into a card—idea, note, meeting, or to-do—sorted for you. Say “the weekly report is in” to check it off; ask “what’s left this week?” for an instant answer.
- ✓Create, complete, reschedule, and search by voice
- ✓To-dos grouped by day, overdue at a glance
Can’t speak?
Type—it still gets it.
In a meeting, late at night, or at a noisy café, double-tap ⌘ and a uTools-style bar appears. Type instead of speaking through the same intent pipeline.
- ✓Summon, use, dismiss—without stealing focus
- ✓Same intent pipeline as voice — consistent results
Still finding the words?
It drafts the next line.
Pause while typing and Moro drafts your next line in light gray, right at the cursor. Tap Tab to take it. Speak when you know what to say; let it draft a sendable line while you're still weighing words — the two ways of writing complement each other.
- ✓Off by default; when enabled it follows your autocomplete model route, with an optional local-only setting; never fires in password fields
- ✓The phrasings you accept become on-device examples, so it fits your voice over time
Everything, in one quiet place
The main window is your overview: today’s activity, the week’s trend, and every recent action at a glance.
Overview
Good evening
Last 24 hours
This week
≈176 min savedQuick actions
Your data,
always under your control
Cloud models are optional and often offer higher quality. One tap switches to local-only, keeping transcription and processing entirely on your Mac.
Bring the models you trust.
Cloud, or fully local
Connect your own API and route per feature, or stay on local models (a free in-app download) and send nothing out.
OpenAI
Anthropic
Gemini
DeepSeek
Qwen
Kimi
Zhipu GLM
Ollama
LM Studio
Mistral
SiliconFlow
MiniMax
MiMo
StepFun
… and any OpenAI / Anthropic-compatible endpoint, plus direct OpenAI Codex subscription sign-in
One purchase, not a monthly bill
Most similar tools are monthly subscriptions. Moro is a one-time purchase with your own models: no cost on-device, and typically just a small amount in the cloud.
Five years of subscriptions cost
—and you keep paying.
~$144/yr ($12/mo) on annual; $15–30/mo monthly · every year.
Pay once for Moro · no usage fee for local models · cloud runs on your own key
Compared against
Prices checked July 2026 and may change; select a name for official pricing.
Built to be a tool you'll want to keep
The best subscription tools are very good after years of refinement. We don't claim to beat them across the board — we focus on the things you use every day, voice, translation, rewriting, notes, and make them smooth, stable, and dependable.
Moro is now in its first private beta
Applications are now open, with participants invited in batches. This one-week beta is for the macOS app; the Windows version is still in development.
First private beta in progress
After applying, download access will be sent by email in batches.
Applications are open. Download access will be sent by email in batches.
Mac first. Windows on the roadmap
Each platform is built natively and polished on its own, without rushing to spread thin.
macOS
Apple Silicon · macOS 15 Sequoia+Feature-complete; final polish and on-device testing underway.
Windows
In developmentThe Windows version is in development and is not included in this beta.
This beta is for macOS only. The Windows version is still in development.
Still have questions?
The ones people ask most, answered up front. Permissions, models, offline and every feature in detail live in the full docs.
Read the full docsHow is it different from system dictation?
System dictation mainly turns speech into raw text. Moro adds intent detection, de-filling, punctuation, tone, translation, notes, Q&A and in-place output so the result is usable. Can't speak? Summon it by typing, or let it draft the next line when you pause (Beta).
Is my data private?
Cloud models are optional. Relevant content is sent to the provider you configure and is subject to that provider’s privacy and retention policies. Moro’s local history is encrypted on your Mac, deletable anytime, and not uploaded by Moro. In local-only mode, transcription and processing can stay entirely on your Mac; password and other secure fields are skipped and not read.
Which languages does it support?
The interface supports Chinese, English, Japanese, and Korean; translation covers more languages with a separate target-language setting. Dictation coverage and accuracy depend on the speech model, accent, and environment. Qwen3-ASR and Fun-ASR support mixed Chinese-English, while Parakeet is better suited to English and European languages.
Which permissions are needed?
Onboarding walks you through four core permissions: Microphone for voice input; Accessibility to power the global shortcuts, read the context at your cursor and write results back; Input Monitoring to keep shortcuts reliable with certain keyboards/IMEs; Full Disk Access for the local “find file” search — names, file contents and recently-opened files are matched against your Mac's own index, nothing leaves the machine. Screen Recording stays off by default and plays no part in everyday input — it's used only if you opt into the experimental reply-draft feature inside closed apps like WeChat.
One-time purchase—what about updates?
The current edition is a one-time purchase for lifetime use, with no monthly or annual renewal. Maintenance updates and future major-version upgrade terms will be published with the final license at launch. Cloud models use your own key and usage; local models have no cloud usage fee.
When can I join the private beta?
The first private beta is now underway. Apply with your email, and download access will be sent by email in batches. Moro requires macOS 15 (Sequoia) or later and is optimized for Apple Silicon.
Moro private beta applications are open
Apply for the first private beta. Download access will be sent by email in batches.
Only for Moro beta invitations, download notices, and product updates.